Detectability of and interference by major and minor hemoglobin variants using a new-generation ion-exchange HPLC system with two switchable analysis modes

Objectives High-performance liquid chromatography (HPLC) is commonly used to measure hemoglobin A1c (HbA1c) levels and detect hemoglobin variants (Hb-Vars). HLC-723GR01 (GR01) is a new-generation automated ion-exchange HPLC system with two switchable analysis modes, namely short (30 s/test) and long modes (50 s/test). We evaluated the general performance of both analysis modes of GR01 for quantifying HbA1c and detecting Hb-Vars. Design and methods We evaluated the instrument's precision based on CLSI protocol EP-05-A3. A comparison of the two analysis modes of GR01 against the standard mode of HLC-723G11 was performed on 100 whole blood samples. The GR01 long mode was compared with affinity HPLC (AF-HPLC) for detecting common Hb-Vars (HbE, HbD, HbS, and HbC, >20 samples). To examine the detection capability for minor Hb-Vars, we analyzed 26 Hb-Vars using multiple analyzers, including both analysis modes of GR01. Results Both modes of GR01 had within-laboratory coefficients of variation of ≤1.0 % from four samples with HbA1c concentrations of 32–86 mmol/mol. Good correlation was observed between GR01 and HLC-723G11. The results for HbA1c detection in the presence of the major variants revealed a strong correlation between the long mode of GR01 and AF-HPLC (r = 0.986–0.998), and the difference biases ranged 0.1–1.9 mmol/mol. In the long mode, only one variant had a difference bias exceeding 14 % [10 % (%NGSP)]. Conclusion The two analysis modes of GR01 were fast and had high accuracy and reproducibility, indicating their utility for routine clinical use in measuring HbA1c samples with Hb-Vars.


Introduction
In 2021, more than 500 million people worldwide had diabetes mellitus, and future projections suggest that the absolute number of people with diabetes will increase by 46 % by 2045 [1].Diabetes is characterized by elevated blood glucose concentrations, and the measurement of hemoglobin A 1c (HbA 1c ) is widely used in the diagnosis of diabetes in defined patient populations and the assessment of glycemic control in patients with diabetes [2].Long-term prospective studies, in particular the Diabetes Control and Complications Trial, the UK Prospective Diabetes Study, and the Kumamoto Study, provided clear evidence that diabetic complications are directly related to the mean blood glucose level as measured by the HbA 1c concentration [3][4][5].Weykamp et al. described HbA 1c as a valuable indicator of long-term glycemic control and defined specific treatment targets and decision limits, and the variable has been successfully standardized [6].
Clinical laboratories use highly accurate methods for HbA 1c testing based on several different methodologies, i.e., cation exchange (CEX) high-performance liquid chromatography (HPLC), affinity HPLC (AF-HPLC), capillary electrophoresis, immunoassay, and enzymatic methods.CEX-HPLC is a highly reliable method for HbA 1c analysis [7].
We developed HLC-723GR01 (GR01) based on CEX-HPLC with two analysis modes for HbA 1c ; a standard short mode (short mode) and a standard long mode (long mode).The short mode measures HbA 1c within 30 s and detects three major hemoglobin variants [Hb-Vars; hemoglobin D (HbD), hemoglobin S (HbS), and hemoglobin C (HbC)] as the H-Var peak and the other major variant [hemoglobin E (HbE)] as the P-HV peak (glycated HbE).The HbA 1c concentration cannot be quantified using the short mode if the H-Var peak is detected.By contrast, the long mode can measure HbA 1c within 50 s and separate the four major Hb-Vars into individual peaks (HbE, P-HV; HbD, D+; HbS, S+; and HbC, C+).When these peaks are detected in the long mode, the HbA 1c concentration is calculated considering each Hb-Var peak, and HbA 1c is reportable.These two modes are switchable using the same column and eluents.
In this study, we evaluated the results of the two analysis modes of GR01 using samples with and without Hb-Vars.

Samples
The ethics committee of Eastern Chiba Medical Center (Chiba, Japan; No. 184-2022) and the Bioscience Division of Tosoh Corporation (Tokyo, Japan; 21-03, 22-03, 22-04) approved this study.At entry, written informed consent was obtained from all participants.
To assess the effect of differences in blood collection tubes on anticoagulants, six patients without diabetes (HbA 1c : 38.0 ± 2.6 mmol/mol) and three donors with diabetes (HbA 1c : 72.Hb-Var samples (HbE, HbD, HbS, HbC, and other rare variants) with HbA 1c levels measured by AF-HPLC measured by Premier Hb9210 (Trinity Biotech, Bray, Ireland) were obtained from the European Reference Laboratory for Glycohemoglobin (Winterswijk, The Netherlands).They were residual samples based on an opt-out or informed consent according to the regulations of each country.The minor Hb-Var samples were genetically tested at Fukuyama Medical Laboratory Co., Ltd.(Hiroshima, Japan) in accordance with the Ethical Guidelines for Genetic Testing.
All clinical investigations were conducted in accordance with the tenets of the Declaration of Helsinki.

HbA 1c analysis
HbA 1c was measured in the short and (30 s/test) and long modes (50 s/test) using GR01 (Tosoh, software ver.1.04) and the standard mode using HLC-723G11 (G11, Tosoh, software ver.3.08).Both GR01 and G11 use CEX-HPLC to separate hemoglobin fractions.Common Hb-Vars can be detected using GR01 in the short mode, but HbA 1c cannot be detected because of interference by Hb-Vars.Conversely, HbA 1c can be detected in the presence of common Hb-Vars using GR01 in the long mode.
All HbA 1c results based on CEX-HPLC were calculated using the total area excluding the hemoglobin F area.HbA 1c values were expressed in the National Glycohemoglobin Standardization Program (NGSP) unit (%, one digit after the decimal point) using calibrators with NGSP units for GR01 evaluations.NGSP units were calculated from International Clinical Federation of Clinical Chemistry and Laboratory Medicine (IFCC) unit (mmol/mol, integer value) using the following formula: NGSP = [0.09148* IFCC] + 2.152 [8].

Statistical analysis
Statistical analyses were performed using Analyse-it (Analyse-it Software, Ltd., Leeds, UK) with Excel (Microsoft Corp., Redmond, WA, USA).

Influence of anticoagulants
To assess whether anticoagulants affect the detection of HbA 1c , blood was collected from donors with and without diabetes into different blood tubes [EDTA2K, EDTA3K, NaF/EDTA2Na (with or without centrifugation), 3.2 % sodium citrate, and heparin lithium].The differences relative to EDTA2K tubes on day 0 were 97%-105 % for the short mode (Supplemental Table 1) and 100%-105 % for the long mode (Supplemental Table 2).No significant tube-dependent effects on HbA 1c measurements were observed using GR01 in the presence of any anticoagulant.The results also indicated that the stability of HbA 1c was not significantly altered in vials containing different anticoagulants after 15 days of storage at 4 • C.

Method comparison between GR01 and G11
A method comparison was performed using 100 fresh whole blood samples from subjects with and without diabetes according to the CLSI EP09-A3 method [10].All results were reported without flags and with normal patterns of chromatograms.Passing-Bablok regression analysis of the data obtained using the short and long modes of GR01 displayed good correlations with the standard mode of G11 (r = 0.999, slope = 1.00, and intercept = 0.0 for both modes; Fig. 1-A and 1-B.Relative to G11, the Bland-Altman plot revealed mean difference biases of 0.26 and 0.37 mmol/mol for the short and long modes of GR01, respectively (Fig. 1-C and 1-D).
In addition, Passing-Bablok regression between the two modes of GR01 revealed a good correlation (r = 0.999) for HbA 1c , as well as a slope of 1.00 and intercept of 0.00, and the Bland-Altman plot revealed a mean different bias of 0.11 mmol/mol (Fig. 1-E and 1-F).

Interference by common Hb-Vars
To evaluate interference by common Hb-Vars in HbA 1c measurement, we performed a method comparison between the long mode of GR01 and the affinity mode of HLC-723G8 using 23 samples containing HbE, 22 samples containing HbD, 21 samples containing HbS, and 23 samples containing HbC.The results of Passing-Bablok regression and Bland-Altman plots are summarized in Table 2 and presented in Fig. 2.

Table 1
Precision testing of GR01 in the short and long modes.Good correlations were observed between the long mode of GR01 and affinity mode of HLC-723G8 using the aforementioned samples (r = 0.986-0.998)with a slope of 0.943-1.000and intercept of 1.484-3.163.

Interference by minor Hb-Vars
We analyzed 26 types of minor Hb-Var and 4 types of common Hb-Var samples using the short and long modes of GR01 and affinity  3, and detailed results obtained using the variant and standard modes of G11, variant modes of HLC-723GX and HLC-723G8, and HPLC based on the KO500 method are shown in Supplementary Figs.1-30.All Hb-Var samples were genetically analyzed by sequencing β-globin, α1-globin, α2-globin, and GAP-PCR (− 3.7, anti 3.7, SEA, FIL).Six (20 %) minor Hb-Vars did not exhibit any flags in the short mode.Only two Hb-Var (7 %) had no flags, and there was no information regarding the detection of the Hb-Var sample even using the long mode.In addition, 16 long and 20 short mode samples had masked HbA 1c concentrations.The HbA 1c levels of one long mode sample (3 %) and three short mode samples (10 %) differed by more than 14 % (mmol/mol) [10 % (% NGSP)] from those of AF-HPLC (Table 3).
The detailed data and chromatograms are presented in Supplemental Figs.1-30.

Discussion
More than 1000 different Hb-Vars have been reported, and the prevalence or types of Hb-Vars differ by race, country, or region [11,  12].Some studies reported the rate of positive populations of Hb-Vars including thalassemia as follows: China, 1074/311,024 (0.35 %) [13]; India, 12,131/65,779 (18.4 %) [14]; United Arab Emirates, 545/6420 (8.5 %) [15]; and Thailand, 636/26,013 (2.4 %) [16].Approximately 20 % of the variants produce clinical phenotypes such as hemolytic anemia, polycythemia, and methemoglobinemia, whereas the remaining 80 % cause no abnormal phenotype [17].Therefore, many of the Hb-Vars themselves do not require routine clinical detection.By contrast, it is critical in HbA 1c testing to account for the presence of Hb-Vars because they sometimes interfere with HbA 1c assessment, and the manner and extent of interference depend on the measurement method [18,19].In addition, some Hb-Vars inhibit red blood cell turnover [20][21][22][23], and the HbA 1c results, regardless of the method of measurement, could deviate from the actual glucose levels.
Recently, Zechmeister et al. reported that HPLC provided more information about the presence of Hb-Vars than enzymatic assays.Analyzers used to measure HbA 1c must have high throughput for rapid testing and high resolution for clearer separation, which represent contradicting goals [24].
To meet these requirements, we developed the new-generation CEX-HPLC analyzer GR01 with two switchable analysis modes, namely short and long, using the same reagents and column.The liquid chromatographic method for HbA 1c measurement was developed by Trivelli et al., in 1971 [25], and the method has been significantly improved.Moreover, HPLC is a potent tool for evaluating Hb-Vars.However, HPLC should not be used as a standalone method for the definitive identification of Hb-Var because some minor Hb-Vars cannot be detected.The accuracy of HbA 1c in principal Hb-Var samples (HbE, HbD, HbS, and HbC) was previously evaluated using commercial methods [26,27].The present study evaluated the performance of GR01 using samples with or without Hb-Vars.The long mode results of GR01 exhibited a good correlation with those of AF-HPLC.Therefore, it was considered that the long mode was free of interference from common Hb-Vars.
In this study, the precision values were consistent with the recommendations of Sacks et al. of within-laboratory CVs of less than 2 % [28].

Table 3
Detection and interference of GR01 in the short and long modes using samples containing minor and common Hb-Vars.HbA 1c levels measured by off-site immunoassay and enzymatic assays have been reported to be significantly lower than those measured by on-site HPLC using blood samples collected in NaF-containing blood collection tubes [29,30].Subsequently, the Committee on Standardization of Laboratory Testing Related to Diabetes Mellitus of the Japan Diabetes Society recommended the use of EDTA-containing tubes for HbA 1c measurement using a layer of erythrocytes harvested by centrifugation [31].We evaluated the effect of anticoagulants on HbA 1c measurements performed using GR01 and blood collection tubes.The results of HbA 1c measurements using either the short or long mode after 15 days were not significantly different from those of the reference EDTA2K tube immediately after blood collection.However, this study used a small number of samples, and the use or storage period of these blood collection tubes was not clarified.The choice of blood collection tubes should be based on the operator's manuals and other information according to the regulations of each country.
As previously mentioned, a large number of Hb-Vars exist globally, making it essential to detect them regardless of whether they interfere with HbA 1c measurements.Previous studies suggested that HbA 1c results should always be interpreted in the clinical context of the patient in case unexpected Hb-Vars are present [18,32].
We evaluated the ability of GR01 in the short and long modes to detect minor Hb-Vars using 26 samples.The results suggested that the presence of Hb-Gouda affected HbA 1c measurements, and the peak could not be clearly determined even by using the long mode.The flag rules need to be improved for the detection of Hb-Vars using a larger number of samples in the future.
This study had two limitations.First, this was a single-center and single-device study.It is desirable to conduct multicenter studies or review multiple evaluation results in the future.The other limitation was that the evaluation was performed using only one sample of each Hb-Var.Therefore, the chromatograms of the same Hb-Vars with different HbA 1c concentrations or with other conditions might be different from those in this report.The 26 types of rare variants used in this study represent only a small portion of the more than 1000 reported Hb-Vars.Therefore, it must be recognized that not all Hb-Vars can be detected, and false reports can occur.
In conclusion, GR01, which has two switchable standard analysis modes, has good reproducibility for rapid HbA 1c measurements.If the frequency of Hb-Vars is low, it might be efficient to use the short mode and retest with the long mode if Hb-Vars are detected using the short mode.However, some Hb-Vars could be missed using either the short or long mode, and caution should be exercised.

Funding
None.

Data statement
The data that support the findings of this study are available from the corresponding author, DM, upon reasonable request.

Declaration of competing interest
DM and SO are employees of Tosoh Corp (Japan).SM is an employee of Tosoh Europe N.V. (Belgium).